Deterministic LLM router with a cryptographic audit trail – TEIA (github.com)

The TEIA Cognitive Router has been introduced as a groundbreaking deterministic LLM router that employs a fixed arithmetic formula to route prompts to different model tiers—Local, Hybrid, or Cloud—while ensuring accountability through a cryptographic SHA-256 audit trail. This router addresses critical compliance needs for organizations under regulations like the EU AI Act and GDPR by providing a mechanism to consistently reproduce routing decisions. Unlike traditional ML-based systems, which can change their routing behavior when weights are retrained, TEIA guarantees that the same input will always yield the same output, enhancing transparency and reliability in AI operations. Significantly, the TEIA Cognitive Router not only establishes a consistent routing framework but also optimizes cost and performance, achieving a throughput of 15,361 prompts per second on modest hardware. It incorporates a Semantic Entropy Score that evaluates prompts based on measurable text features, allowing for nuanced routing decisions with a lower operational cost—reporting a projected 16.3% cost reduction compared to a fully cloud-based model. The inclusion of cryptographic verification and a standalone verifier further fortifies the integrity of AI-driven processes, making TEIA a vital tool for organizations striving for both efficiency and compliance in the rapidly evolving AI/ML landscape.